Deep insights into gastrointestinal health: A comprehensive analysis of GastroVision dataset using convolutional neural networks and explainable AI

Muhammad Fahad, Noor E Mobeen, Ali Shariq Imran, Sher Muhammad Daudpota, Zenun Kastrati, Faouzi Alaya Cheikh, Mohib Ullah

Abstract: Highlights•Proposed a novel approach that leverages Greedy Soup with DenseNet-121 for GI disease detection.•Performed SoTA experiments on the GastroVision dataset and with dataset categories.•Enhanced accuracy using GAN and data augmentation.•Incorporated XAI (SHAP and GradCAM) for the model interpretability.

External IDs:doi:10.1016/j.bspc.2024.107260